/**
* Get index
* @return \ZendSearch\Lucene\Index
*/
private function index()
{
if (!isset(self::$index)) {
$analyzer = new CaseInsensitive();
if ($this->config()->exists('zend_search', 'stop_words')) {
$stop_word_filter = new StopWords();
$words = $this->getRealPath($this->config()->get('zend_search', 'stop_words'));
if ($words !== false) {
$stop_word_filter->loadFromFile($words);
} else {
throw new \InvalidArgumentException('Path not found');
}
$analyzer->addFilter($stop_word_filter);
}
if ($this->config()->exists('zend_search', 'morphy_dicts')) {
$morphy_dicts = $this->getRealPath($this->config()->get('zend_search', 'morphy_dicts'));
if ($morphy_dicts !== false) {
$analyzer->addFilter(new Morphy($morphy_dicts, $this->config()->getCharset()));
} else {
throw new \InvalidArgumentException('Path not found');
}
}
Analyzer::setDefault($analyzer);
Lucene::setResultSetLimit($this->limit);
QueryParser::setDefaultEncoding($this->config()->getCharset());
$index = $this->config() - get('zend_search', 'index');
$path = $this->getRealPath($index);
self::$index = $path ? Lucene::open($path) : Lucene::create($index);
}
return self::$index;
}
public function testUtf8CaseInsensitive()
{
if (@preg_match('/\\pL/u', 'a') != 1) {
// PCRE unicode support is turned off
return;
}
if (!function_exists('mb_strtolower')) {
// mbstring extension is disabled
return;
}
/** Zend_Search_Lucene_Analysis_Analyzer_Common_Utf8_CaseInsensitive */
$analyzer = new Utf8\CaseInsensitive();
// UTF-8 text with a cyrillic symbols
$tokenList = $analyzer->tokenize('Слово1 Слово2 ДругоеСлово', 'UTF-8');
$this->assertEquals(count($tokenList), 3);
$this->assertEquals($tokenList[0]->getTermText(), 'слово');
$this->assertEquals($tokenList[0]->getStartOffset(), 0);
$this->assertEquals($tokenList[0]->getEndOffset(), 5);
$this->assertEquals($tokenList[0]->getPositionIncrement(), 1);
$this->assertEquals($tokenList[1]->getTermText(), 'слово');
$this->assertEquals($tokenList[1]->getStartOffset(), 7);
$this->assertEquals($tokenList[1]->getEndOffset(), 12);
$this->assertEquals($tokenList[1]->getPositionIncrement(), 1);
$this->assertEquals($tokenList[2]->getTermText(), 'другоеслово');
$this->assertEquals($tokenList[2]->getStartOffset(), 14);
$this->assertEquals($tokenList[2]->getEndOffset(), 25);
$this->assertEquals($tokenList[2]->getPositionIncrement(), 1);
}
